A recent study soon to be published in the

*Journal of Educational Psychology*attempts to determine analytically how homework correlates with learning outcomes. The study involves more than 7,000 students in their second year of obligatory secondary education in the principality of Asturias in Spain. The subjects examined in this work are mathematics and science. Data on homework are collected and scores from standardized tests in mathematics and science are used as measures for learning outcomes. Teachers in these schools generally assign homework that is about 70 minutes long per day. The data indicate that the amount of homework is generally uniform among teachers. However, there is an observed variance in the times reported by students. This observed variability therefore comes mainly from the students themselves, specifically their commitment to do homework. With the differences in time spent on homework, a preliminary analysis on how time spent on homework is related to test scores can be made (shown in the following figure):

In both subjects, the relationship is a curve. This suggests that there is an optimum time spent on homework (around 90-100 minutes per day). In addition, there are diminishing returns. For example, the gain in score per minute of homework is larger near 50 minutes per day than at times near the optimum. Thus, adding 20 minutes of homework to a child who does 50 minutes a day may lead to a much more significant difference than to a child who is already spending 70 minutes a day.

Aside from the amount of time, homework likewise comes in different flavors of frequency. The study shows that learning outcomes correlate positively with how often homework is assigned. Therefore, although more homework does not always correlate with better results especially when it takes too much time, more frequent homework always goes with better scores. Frequent but not long therefore seems to be key.

Applying more rigorous analysis of the data by introducing other factors such as amount of effort and autonomy, socio-economic status, gender, and prior knowledge reveals a lot more about how homework may in fact be affecting learning. By taking into consideration how a student does the homework (measuring a student's effort and self-reliance) erases the correlation between learning outcomes and homework time. Hence, what counts is not the time but how much effort and independence a child exerts in doing the homework. Incorporation of the educational attainment and professions of the students’ parents likewise explains a significant amount of variance in the learning outcomes. However, all of these factors become much less significant when prior grades of students in mathematics and science are included in the analysis. Effort no longer accounts but autonomy is still important. Socio-economic factors become half as important as well.

Students who have had good grades in math and science previously no longer have their learning outcomes depend as much as other students do on time spent on homework, effort, and socio-economic status. The fact that autonomy remains consequential probably provides a clue on what role homework really serves in a students' learning. The authors of the study write:

Homework is not a mere extension of the classroom. It is not an augmentation of teaching or learning time. Instead, homework is an opportunity to develop responsibility and self-efficacy. With this in mind, it becomes clear what the real value of homework is in education. It likewise becomes evident what a good homework should be.Our data indicate that it is not necessary to assign huge quantities of homework, but it is important that assignment is systematic and regular, with the aim of instilling work habits and promoting autonomous, self-directed learning.

The above study likewise reinforces the message from a previous post in this blog which discusses homework in the early years of education.

*A child's reading achievement is facilitated by parental involvement in homework through an improvement in the child's academic functioning (more effort and less procrastination)*.